Receiver equalizer circuitry with offset voltage compensation for use on integrated circuits

BACKGROUND

This invention relates to receiver equalizer circuitry, and more particularly to receiver equalizer circuitry having offset voltage compensation circuitry. The equalizer circuitry is intended for implementation on an integrated circuit (“IC”), especially an IC fabricated using, for example, 28 nm CMOS technology. The equalizer circuitry is further intended for use in receiving and equalizing (i.e., improving the quality of, and therefore the ability to recover data information from) a high-speed serial data signal. For example, the signal to be equalized may have a bit rate in the range of about 20-25 Gigabits-per-second (“Gbps”).

SUMMARY

In accordance with certain possible aspects of the invention, equalizer circuitry on an IC may include first, second, and third continuous time, linear, equalizer stages connected in series with one another. Each of the equalizer stages preferably includes peaking inductor circuitry. The equalizer circuitry may further include controllably variable, static, DC mode offset voltage compensation circuitry for controllably reducing DC voltage offset between an output of the third equalizer stage and utilization circuitry (e.g., clock and data recovery (“CDR”) circuitry) to which that output is applied. As an alternative or addition to the above static offset voltage compensation circuitry, the equalizer circuitry may include dynamic, continuous mode, DC offset voltage compensation circuitry for automatically reducing, on an on-going or continuous basis, DC voltage offset between the third stage output and the utilization circuitry.

In accordance with another possible aspect of the invention, the first equalizer stage may be preceded by termination circuitry having controllably variable impedance.

In accordance with still another possible aspect of the invention, differential circuitry and differential signalling may be used for various components of the equalizer circuitry.

Further features of the invention, its nature and various advantages will be more apparent from the accompanying drawings and the following detailed description.

BRIEF DESCRIPTION OF THE DRAWINGS

<figref idrefs="DRAWINGS">FIG. 1</figref> is a simplified, schematic, block diagram of an illustrative embodiment of certain possible features in accordance with the invention.

<figref idrefs="DRAWINGS">FIG. 2</figref> is a simplified, schematic, block diagram of an illustrative embodiment of a representative portion of the <figref idrefs="DRAWINGS">FIG. 1</figref> circuitry in accordance with certain possible features of the invention.

<figref idrefs="DRAWINGS">FIG. 3</figref> is a simplified, schematic, block diagram of an illustrative embodiment of another portion of the <figref idrefs="DRAWINGS">FIG. 1</figref> circuitry in accordance with certain other possible features of the invention.

<figref idrefs="DRAWINGS">FIG. 4</figref> is a simplified, schematic, block diagram of an illustrative embodiment of a portion of the <figref idrefs="DRAWINGS">FIG. 2</figref> circuitry in accordance with certain still other possible features of the invention.

<figref idrefs="DRAWINGS">FIG. 5</figref> is similar to <figref idrefs="DRAWINGS">FIG. 4</figref> for yet other possible features of the invention.

<figref idrefs="DRAWINGS">FIG. 6</figref> is similar to <figref idrefs="DRAWINGS">FIG. 2</figref>, but shows an illustrative embodiment of modification of one instance of the <figref idrefs="DRAWINGS">FIG. 2</figref> circuitry in accordance with certain possible aspects of the invention.

<figref idrefs="DRAWINGS">FIG. 7</figref> is similar to <figref idrefs="DRAWINGS">FIG. 6</figref> for another instance of the <figref idrefs="DRAWINGS">FIG. 2</figref> circuitry in accordance with certain other possible aspects of the invention.

DETAILED DESCRIPTION

An illustrative embodiment of a receiver equalizer <b>10</b> in accordance with this invention is shown in <figref idrefs="DRAWINGS">FIG. 1</figref>. This is an inductor-based peaking equalizer that can be used, for example, for 20-25 Gigabits-per-second (“Gbps”) data signals on a relatively short-range data link. (An illustrative embodiment of one representative continuous time, linear, equalizer (“CTLE”) stage <b>30</b> with peaking inductors that can be used in equalizer <b>10</b> is shown in <figref idrefs="DRAWINGS">FIG. 2</figref>.) Equalizer <b>10</b> can be illustratively implemented in an integrated circuit (“IC”) using 28 nanometer (“nm”) CMOS technology as part of a high-capacity programmable logic device (“HCPLD”) or field-programmable gate array (“FPGA”), although other uses (e.g., in programmable microcontrollers) are also possible. As used herein, the term “complex programmable logic device” (“CPLD”) will sometimes be used as a generic term for all of these types of integrated circuits (e.g., HCPLDs, FPGAs, programmable microcontrollers, etc.).

As process technologies scale down, the intrinsic speed (ft) of CMOS transistors moves to higher frequencies. This benefits high-frequency design at lower total power. However, 28 nm CMOS technology is still evolving, so that some process parameters are not yet finalized or are still in development. Also, high gate resistance and high-K metal gate issues may pose additional challenges to the stability of this technology in the short term. Parasitic capacitances and resistances have increased compared to other deep sub-micron technology. These factors increase the design challenges that are faced in trying to provide enough bandwidth to overcome the loading limits.

Traditional peaking inductors manage to get higher bandwidth, but at the price of very large inductor area. This is avoided in the present design by using (for example) a 1 nH inductor having a much smaller area. In particular, a solenoid inductor (as an example) can reduce intra-winding capacitance, thereby enhancing the Q for a given area, while also achieving higher inductance. This may permit a 6-10 times reduction in area compared to traditional spiral inductors. The solenoid inductors thus referred to (and preferred for use in accordance with this invention) are of a type that is per se known. Such an inductor is fully integrated into the IC. It may employ multiple (e.g., three) metal layers, rather than one metal layer, to save area. The turns of the inductor spiral vertically down through the multiple layers used to provide the inductor. It provides high Q, and has as another important advantage a high self-resonance frequency because the coupling between turns is reduced. Thus it is a kind of spiral inductor, but the spiral is vertical, not horizontal or flattened.

As will be described below, equalizer <b>10</b> preferably includes voltage offset compensation circuitry to correct process/voltage/temperature (“PVT”) mismatch that may occur. Without such offset compensation, PVT mismatch could make equalizer circuitry <b>10</b> not an ideal interface between the input data signals applied to the equalizer and the circuitry (e.g., clock and data recovery or “CDR” circuitry) downstream from the equalizer. To address this possible problem, equalizer <b>10</b> preferably includes both static calibration offset and continuous mode DC-offset correction with negative feedback. (A generic term that is sometimes used for the downstream circuitry that makes use of the output of equalizer <b>10</b> is “utilization circuitry,” CDR circuitry being an illustrative example of such utilization circuitry.)

As has already been said, equalizer <b>10</b> can be implemented, for example, in 28 nm CMOS technology to provide, for example again, 25 Gbps physical layer implementation. Compared with ten 10 Gbps lanes, only four channels of such 25 Gbps circuitry can meet the IEEE802.3ba specification for 100 Gbps Ethernet with less power consumption and area cost. This receiver equalizer also allows support of backplane applications, which increases the number of possible uses of ICs that include this circuitry.

As shown in <figref idrefs="DRAWINGS">FIG. 1</figref>, equalizer <b>10</b> includes four main blocks. These are termination block <b>12</b>, equalizer block <b>14</b>, dynamic (continuous mode) voltage offset compensation block <b>16</b>, and static voltage offset compensation block <b>18</b>.

A received serial data signal to be processed by equalizer <b>10</b> (and thereafter by downstream utilization circuitry (e.g., CDR circuitry) that is not shown) is applied to termination block <b>12</b> in differential from (i.e., as two complementary signals RXP and RXN). Termination block <b>12</b> is differential circuitry that includes two resistors <b>20</b><i>a </i>and <b>20</b><i>b </i>that are connected in series between the leads carrying the RXP and RXN signals. A source of common mode voltage VCM is connected to a node between resistors <b>20</b><i>a </i>and <b>20</b><i>b. </i>

A typical objective is for termination block <b>12</b> to generate a 100-ohm DC differential impedance. To help ensure that there always is the desired (e.g., 100-ohm) impedance, termination block <b>12</b> is preferably tunable. For example, <figref idrefs="DRAWINGS">FIG. 3</figref> shows that resistors <b>20</b><i>a </i>and <b>20</b><i>b </i>may have controllably variable and selectable resistance(s). In particular, <figref idrefs="DRAWINGS">FIG. 3</figref> shows that each of resistors <b>20</b><i>a </i>and <b>20</b><i>b </i>may comprise respective resistors R<b>1</b>, R<b>2</b>, and R<b>3</b>, with resistors R<b>1</b> and R<b>2</b> being fixed resistors connected in series with one another, and with resistor R<b>3</b> being a controllably variable resistor connected in parallel with the associated resistor R<b>2</b>.

The resistance of each resistor R<b>3</b> is controlled by the output signal(s) of associated multiplexer (“mux”) circuitry <b>120</b><i>a </i>or <b>120</b><i>b</i>. Each mux can select its output(s) from either an associated static (or at least relatively static) signal source <b>122</b><i>a </i>or <b>122</b><i>b</i>, or alternatively from a more dynamic signal source. The selection made by each mux <b>120</b> is determined by the logical state of selection control element <b>124</b>. For example, if equalizer <b>10</b> is implemented on a programmable logic device (“PLD”), a complex PLD (“CPLD”), or other type of programmable integrated circuit device, static sources <b>122</b> and <b>124</b> may be random access memory (“RAM”) bits that are provided on the IC to initially “program” or “configure” (or reprogram or reconfigure) the IC for its desired subsequent “normal mode” or “user mode” operations. In such cases, the more dynamic signals referred to in <figref idrefs="DRAWINGS">FIG. 3</figref> may be time-varying signals that occur on the IC during its normal or user mode operations. Although more dynamic than configuration RAM (“CRAM”), these dynamic signals may be stored in non-CRAM memory or registers on the IC. The output signals of each mux <b>120</b> may be three binary digits (“bits”), providing three-bit control of the amount of resistance provided by the associated variable resistor R<b>3</b>, and hence the associated resistor <b>20</b>.

The common mode voltage VCM may also be selectable to meet the requirements of any of several possible applications of the circuitry. For example, VCM may be selected (e.g., by CRAM control) to be 0.7V, 0.65V, or any other desired common mode voltage level.

With regard to tuning termination block <b>12</b> as described above, there is a tradeoff between the need to compensate for process variations and a need to avoid undue parasitic capacitance.

Turning now to equalizer block <b>14</b>, this is a three-stage, differential, analog, linear equalizer with peaking inductors. The three stages are labelled EQ<b>1</b>, EQ<b>2</b>, and EQ<b>3</b> (or <b>30</b><i>a</i>, <b>30</b><i>b</i>, and <b>30</b><i>c</i>), and they are connected in series in that order. Each stage <b>30</b> may be constructed basically as shown in <figref idrefs="DRAWINGS">FIG. 2</figref> (although EQ<b>1</b> and EQ<b>2</b> preferably have some additional features that will be described after the general discussion of <figref idrefs="DRAWINGS">FIG. 2</figref> that follows first). With three stages <b>30</b><i>a</i>-<i>c</i>, equalizer block <b>14</b> can provide, for example, 9-dB gain at half data rate frequency, and up to 15-dB gain by appropriately tuning the Rz and Cz values in the three stages (see later discussion of <figref idrefs="DRAWINGS">FIG. 2</figref>). As noted earlier, the equalizer is particularly adapted to compensate distances of less than, e.g., 20 cm link length including connectors and vias. As was also noted earlier, the peaking inductor value may be about 1 nH with a Q of about 5 at the frequencies of interest.

<figref idrefs="DRAWINGS">FIG. 2</figref> shows the basic construction of each equalizer stage <b>30</b> in <figref idrefs="DRAWINGS">FIG. 1</figref>. The differential circuitry of each stage <b>30</b> is connected between a power supply voltage VDD and ground. The circuitry basically includes two parallel, differential legs, each of which includes a peaking inductor <b>32</b><i>a </i>or <b>32</b><i>b</i>, a resistor <b>34</b><i>a </i>or <b>34</b><i>b</i>, source and drain terminals of a transistor <b>36</b><i>a </i>or <b>36</b><i>b</i>, and a current source <b>38</b><i>a </i>or <b>38</b><i>b</i>. In each leg, the above circuit elements are connected in series in the order in which they are named above. The input data signal (to be equalized or further equalized) is applied in complementary (differential) form to input terminals INP and INN, which are respectively connected to the gates of transistors <b>36</b><i>a </i>and <b>36</b><i>b</i>. These complementary signals are either the outputs of termination block <b>12</b> or the outputs OUTP and OUTN of the immediately preceding equalizer block <b>30</b>, depending on which stage of the equalizer is being considered. The equalized data output signal of the equalizer stage is applied in complementary or differential form to output leads OUTN and OUTP, which are connected to nodes between resistors <b>34</b> and transistors <b>36</b>. (Capacitors <b>40</b><i>a </i>and <b>40</b><i>b </i>(or C<b>1</b>) represent the loading capacitance seen at the output nodes. For example, such loading capacitance may include parasitic loading from the routing, and the device loading that connects to the node (e.g., gate capacitance of the next stage input devices).) Variable capacitor <b>40</b> (or Cz) is connected between (1) a node between transistor <b>36</b><i>a </i>and current source <b>38</b><i>a</i>, and (2) a node between transistor <b>36</b><i>b </i>and current source <b>38</b><i>b</i>. Variable resistor <b>50</b> (or Rz) is connected in parallel with capacitor <b>40</b> between immediately above-mentioned nodes (1) and (2).

In order to optimize the equalizer for different environments, equalizer stage <b>30</b> provides certain tuning ability. In particular, the capacitor <b>40</b> (typically implemented as a varactor) and resistor <b>50</b> values (Cz and Rz, respectively) have certain tuning ranges that can be controlled, for example, by the CPLD core (e.g., in cases in which the equalizer is implemented on that type of IC). Such tuning can adjust the pole and zero locations of the equalizer stage to optimize the equalizer curve for various transmission environments. Integrating the equalizer circuitry with CPLD circuitry allows programming or controlling the different tuning settings quickly and at low cost to the user.

<figref idrefs="DRAWINGS">FIGS. 4 and 5</figref> show examples of how variable capacitor <b>40</b> and variable resistor <b>50</b> may be controlled (tuned) as described above. The control circuitry shown in <figref idrefs="DRAWINGS">FIGS. 4 and 5</figref> is generally of the type shown in <figref idrefs="DRAWINGS">FIG. 3</figref>. Thus in <figref idrefs="DRAWINGS">FIG. 4</figref>, the capacitance value Cz of variable capacitor <b>40</b> may be controlled by the output(s) of mux <b>140</b>. Mux <b>140</b> selects its outputs from either static or relatively static source <b>142</b> (e.g., configuration RAM or CRAM bits) or from more dynamic signals (e.g., user mode signals or user RAM of associated CPLD core circuitry). The selection made by mux <b>140</b> is controlled by a selection control signal from source <b>144</b> (e.g., another CRAM bit). The circuitry shown in <figref idrefs="DRAWINGS">FIG. 5</figref> is similar, with <figref idrefs="DRAWINGS">FIG. 5</figref> elements <b>150</b>, <b>152</b>, <b>154</b>, and “dynamic signal(s)” being respectively similar to <figref idrefs="DRAWINGS">FIG. 4</figref> elements <b>140</b>, <b>142</b>, <b>144</b>, and “dynamic signal(s).”

Turning now to offset voltage compensation circuits <b>16</b> and <b>18</b> in <figref idrefs="DRAWINGS">FIG. 1</figref>, such offset voltage compensation is important to reduce process-variation-caused mismatch (also sometimes referred to as PVT mismatch), especially for geometries below 90 nm technology. The present disclosure preferably includes both static DC mode offset voltage compensation <b>18</b> and dynamic continuous mode offset voltage compensation <b>16</b>. Static compensation <b>18</b> calibrates the DC voltage offset from the equalizer output to, for example, the phase detector of downstream CDR circuitry (not shown) during initial power-up of the circuitry (e.g., the CPLD) that includes the equalizer. It provides what may be described as relatively coarse compensation, e.g., up to about 80 mV with about 5 mV resolution. For example, CPLD core circuitry may provide (e.g., store or program) eight bits to calibrate the appropriate static DC mode offset voltage. This static DC mode offset compensation may be actually effected in the second stage <b>30</b><i>b </i>of equalizer <b>14</b>, as will be shown and described in more detail later in this specification.

The continuous mode offset voltage compensation provides relatively fine compensation while the equalizer is in normal operation. This is the circuitry shown at <b>16</b> in <figref idrefs="DRAWINGS">FIG. 1</figref>. This differential circuitry monitors the differential voltage output of equalizer <b>14</b> (i.e., the differential voltage of differential signal CDRP/CDRN), and dynamically compensates for any low-frequency or DC offsets. In other words, whereas the offset compensation provided by the earlier-mentioned static DC mode offset compensation circuitry <b>18</b> is basically constant after it has been established on the basis of an initial calibration of the IC prior to normal-mode use of the IC, the amount of compensation provided by circuitry <b>16</b> can vary over time during the course of normal-mode use of the IC. Moreover, this dynamic variation of the compensating effects of circuitry <b>16</b> occurs automatically in response to on-going monitoring of the amount and direction of offset that needs to be corrected (compensated).

Continuous mode offset voltage compensation circuitry <b>16</b> includes error amplifier circuit A<b>1</b> (or <b>70</b>) and so-called Gm stage <b>90</b>. Circuit <b>70</b> may be a so-called fully differential, folded-cascode, trans-conductance amplifier with common mode feedback (“CMFB”), e.g., of a type that is per se conventional and well known to those skilled in the art. It typically does not have internal tuning. Gm <b>90</b> can be a relatively simple differential circuit that converts differential voltage offset error signals output by A<b>1</b> to proportional differential currents (In and Ip) with low gain. Gm <b>90</b> may be programmable (e.g., using CRAM) to provide any of several desired current gains.

Resistors Rc are connected in series between (1) CDRP and CDRN, and (2) the differential inputs to A<b>1</b>. Cp indicates parasitic capacitance seen at these nodes. Such parasitic capacitance typically includes the routing parasitic capacitance and input device gate capacitance. Each capacitor Co is connected between (1) a respective one of the differential outputs of A<b>1</b>, and (2) ground. Ro refers to the output impedance of the error amplifier. The differential outputs of A<b>1</b> are connected to differential inputs to Gm. Gm produces two, typically dynamically time-varying output currents In and Ip that are used in first equalizer stage <b>30</b><i>a </i>as shown more specifically in <figref idrefs="DRAWINGS">FIG. 6</figref> and described later.

Rc, Cp, Ro, and Co define the pole positions of dynamic continuous mode offset compensation circuitry <b>16</b>. The locations of these poles influence the stability of this closed loop feedback offset compensation control. Rc and Co are variable to provide some tunability to compensate for process variations. For example, the circuitry that is used to control the variable capacitance of each of capacitors Co can be constructed similarly to what is shown in <figref idrefs="DRAWINGS">FIG. 4</figref> for earlier-described variable capacitors <b>40</b>. Likewise, the circuitry that is used to control the variable resistance of each of resistors Rc can be constructed similarly to what is shown in <figref idrefs="DRAWINGS">FIG. 5</figref> for earlier-described variable resistors <b>50</b>. Again, Ro is the output impedance of A<b>1</b>, and Cp is the capacitance at A<b>1</b>'s inputs.

The bandwidth of dynamic continuous mode offset compensation loop <b>16</b> is preferably set at a low frequency (e.g., less than 100 MHz) to avoid interfering with the higher-frequency data path (i.e., from RXP/RXN to CDRP/CDRN).

<figref idrefs="DRAWINGS">FIG. 6</figref> shows how the output current In and Ip of offset compensation circuit <b>16</b> may be used in equalizer block <b>14</b> to actually produce dynamic continuous mode offset compensation. In particular, <figref idrefs="DRAWINGS">FIG. 6</figref> shows that In (output by circuit block <b>14</b>) is connected in parallel with current source <b>38</b><i>a </i>in first-stage equalizer <b>30</b><i>a </i>so that In and the current of source <b>38</b><i>a </i>algebraically add together. Similarly, <figref idrefs="DRAWINGS">FIG. 6</figref> shows that Ip (output by circuit block <b>14</b>) is connected in parallel with current source <b>38</b><i>b </i>in first-stage equalizer <b>30</b><i>a </i>so that Ip and the current of source <b>38</b><i>b </i>algebraically add together. The offset compensation shown here does not load the high-speed outputs of the equalizer stage. It therefore comes with little penalty to the performance of the equalizer.

<figref idrefs="DRAWINGS">FIG. 7</figref> shows an illustrative embodiment of circuitry that may be used, e.g., in second equalizer stage <b>30</b><i>b</i>, to effect static DC mode compensation <b>18</b> as described earlier in this specification. In particular, <figref idrefs="DRAWINGS">FIG. 7</figref> shows that in second stage <b>30</b><i>b </i>variable current source <b>138</b><i>a </i>is connected in parallel with current source <b>38</b><i>a </i>so that the currents of these sources algebraically add together. Similarly, <figref idrefs="DRAWINGS">FIG. 7</figref> shows that second stage variable current source <b>138</b><i>b </i>is connected in parallel with second stage current source <b>38</b><i>b </i>so that the currents of these two sources algebraically add together. The current In<b>2</b> produced by variable current source <b>138</b><i>a </i>is controlled by memory or registers (RAM bits) <b>238</b><i>a</i>. The current Ip<b>2</b> produced by variable current source <b>138</b><i>b </i>is controlled by memory or registers (RAM bits) <b>238</b><i>b</i>. Elements <b>138</b><i>a/b </i>and <b>238</b><i>a/b </i>can constitute the static offset circuitry <b>18</b> shown in <figref idrefs="DRAWINGS">FIG. 1</figref>.

The contents of registers <b>238</b><i>a/b </i>are set as part of a static DC mode calibration operation that may take place (as mentioned earlier in this specification) during initial power up of the IC. During this process, there is no high-speed serial data RXP/RXN, first equalizer stage <b>30</b><i>a </i>is powered down, dynamic mode compensation block <b>16</b> is not operating, and the inputs to second equalizer stage <b>30</b><i>b </i>are set to the same common mode voltage. As a result, the second and third stage equalizers (<b>30</b><i>b </i>and <b>30</b><i>c</i>) are both on with input common mode voltage. The offset voltage compensation value needed is detected via test circuitry and converted to appropriate values for storage in registers <b>238</b> by appropriate external analysis (e.g., a look-up table that converts the DC voltage offset from equalizer block <b>14</b> to the phase detector of downstream CDR circuitry to the RAM bit <b>238</b> values needed to eliminate or at least substantially reduce the detected offset). Registers <b>238</b> are loaded with the RAM bit values determined in this way, and thereafter blocks <b>14</b> and <b>16</b> are allowed to begin normal operation.

By combining the above-described static and dynamic modes, the voltage offset compensation range can be quite large (e.g., up to 130 mV). The minimum compensated offset voltage is limited by error amplifier <b>70</b>, since it also introduces offset voltage. To minimize the offset voltage, the input differential pairs (at the gates of input transistors of A<b>1</b> (<b>70</b>)) should be big enough. These input transistor gates are connected to CDRP and CDRN through resistors Rc. The values of Rc are preferably selected to be big enough to minimize the loading penalty to CDRP/CDRN from the A<b>1</b> input transistors. Thus the values of Rc are a trade-off among stability, area, and equalizer performance. The loop stability is defined by the error amplifier <b>70</b> output stage pole and input pole position. The loading capacitor Co should be big enough to keep more than 60 degrees phase margin (“PM”) for the PVT. All of the circuits are preferably fully differential to have better common-mode noise rejection.

Briefly recapitulating and in some respects extending the foregoing, at high frequencies equalizer <b>10</b> is important to recover the data in the receiver because of the bandwidth limitation. This design is a novel receiver equalizer at frequencies, e.g., of 25 Gbps in, e.g., 28 nm CMOS technology with low power supply of, e.g., 1.0V. With a three-stage analog linear equalizer (instead of, e.g., a decision feedback equalizer (“DFE”)), the three-stage equalizer consumes, e.g., about 16.5 mA DC current with an, e.g., 1.0V power supply to open, e.g., 25 Gbps PRBS7 data through an, e.g., 300 mm channel with, e.g., 16 dB of loss at, e.g., 12.5 GHz. With combined static and dynamic offset compensation, it permits operation with longer links up to, e.g., 25 cm without significant calibration and tuning. Integrating the equalizer into an IC such as a CPLD or a high-end field programmable gate array (“FPGA”) increases the flexibility of high-speed data links. As mentioned, the equalizer can be fabricated with, e.g., 28 nm CMOS technology running at, e.g., 25 Gbps and integrated into, e.g., an HCPLD. This helps make possible, e.g., 20/25 Gbps backplane applications.

It will be understood that the foregoing is only illustrative of the principles of the invention, and that various modifications can be made by those skilled in the art without departing from the scope and spirit of the invention.